Many companies install and maintain search software for use on a variety of distributed systems that can range from a single computer for a small business to a collection of servers and a plurality of user computer nodes for a large corporation. In the past, search reports based on relatively large sets of indexed data were time consuming to generate. To reduce latency in search report generation for a larger data set intermediate summaries for reoccurring search reports for partitioned data have been periodically generated and stored in a separate index. Subsequently, a search report run on the data set would aggregate the corresponding pre-computed intermediate summaries to generate the search report in a relatively shorter period of time. However, the nature of different types of data and the structure of different search queries has made it difficult to manage and configure these intermediate summaries. For example, some partitions may include data that is highly reducible for a search query and other partitions may include data that is only marginally reducible for the same search query. Also, for a single set of data in a partition, it may be reducible for some queries but not others. Other difficulties include configuring the amount of data included in the partitions of the data set, such as adding newly identified data to a partition after intermediate summaries are generated. Consequently, systems that can manage search report generation for relatively large sets of data are the subject of considerable innovation.